Payment Card Rewards Programs and Consumer Payment Choice Abstract

Payment Card Rewards Programs and Consumer Payment Choice
Andrew Ching *
University of Toronto
Fumiko Hayashi **
Federal Reserve Bank of Kansas City
July, 2006
Preliminary. Please do not cite or quote without permission.
Abstract
Card payments have been growing very rapidly. To continue the growth, payment card networks keep
adding new merchants and card issuers try to stimulate their existing customers’ card usage by providing
rewards. This paper seeks to analyze the effects of payment card rewards programs on consumer payment
choice, by using consumer survey data. Specifically, we examine whether credit/debit reward receivers
use credit/debit cards relatively more often than other consumers, if so how much more often, and which
payment methods are replaced by reward card payments. Our results suggest that (i) consumers with
credit card rewards use credit cards much more exclusively than those without credit card rewards; (ii)
even among those who carry a credit card balance, consumers with credit card rewards use a credit card
more often than those without rewards; (iii) among consumers who receive credit card rewards, those who
receive credit card rewards as well as debit card rewards tend to use debit cards more often than those
who receive credit card rewards only; and (iv) reward card transactions seem to replace not only paperbased transactions but also non-reward card transactions.
*
Rotman School of Management, University of Toronto, [email protected]
Payments System Research, Federal Reserve Bank of Kansas City, [email protected]. The views
expressed in this paper are those of the authors and do not necessarily reflect those of the Federal Reserve Bank of
Kansas City or the Federal Reserve System.
**
1.
Introduction
Credit and debit card payments have been growing very rapidly. Debit card outpaced
credit card in terms of number of transactions in 2003, while credit card annual transaction value
was still twice as much as debit card annual transaction value in 2004. To continue the growth,
payment card networks keep adding new merchants to their networks. 1 Penetrating new
cardholders, on the other hand, is becoming difficult because most consumers have both credit
and debit cards. 2 Payment card issuers, therefore, are trying to stimulate their existing customers’
card usage by providing rewards. 3 It has been reported that many credit and debit card issuers
that launched new rewards programs have seen increases in spending on their cards. 4 However,
we are not aware of any reports telling the sources of these increases. It is unlikely that reward
receivers simply increase their spending on their card without changing spending through other
payment methods. Which payment methods are replaced by reward card transactions? Do reward
card transactions replace transactions of other payment methods, such as cash, checks, and/or
ACH? Or do they replace non-reward card transactions? To what extent do reward card
transactions replace other forms of payment transactions?
The answers of above questions are important to policymakers. It is cost effective if
reward card transactions replaced other types of transactions which are more costly than reward
card transactions. It is not cost effective, however, if reward card transactions replaced nonreward card transactions. Operating a rewards program is not free—it uses some resources.
Another concern is that rewards credit cards could potentially create inequality among
1
Credit and debit card adoption rates vary by industry. The rate is close to 100% in some industries.
According to the 2001 Survey of Consumer Finance (SCF) conducted by the Federal Reserve, 75 percent of U.S.
households hold at least one credit card. Statistics on debit card penetration is hard to obtain, but many large banks
reported that 80 to 100 percent of their checking account holders hold a debit card.
3
Credit card rewards programs have a long history. However, it is only recently that issuers started offering more
generous rewards programs. Recently, debit card rewards programs are also getting popular.
2
2
consumers. Many merchants need to pay higher fees to issuers if their customers use a reward
credit card instead of using a non-reward credit card. 5 Credit card networks do not allow
merchants to reject reward card payments as long as the merchants accept the network’s nonreward credit cards. The networks also prohibit merchants from price discriminating customers
based on the payment method they use. 6 As a result, the more customers use reward credit cards,
the higher the merchants mark up their uniform retail prices (in order to offset higher fees).
Although reward credit card holders are partly compensated for higher retail prices through
rewards, other consumers are not. Furthermore, it should be noted that reward credit card holders
are relatively high-income earners, while many low income customers may not even qualify for
having credit cards. 7 Therefore, rewards programs and the accompanied merchant fee structure
may work as tools that distribute income from low-income earners to high-income earners.
Understanding how rewards programs change the market share of other payment methods
is also important to card issuers. For example, it may not be profitable for an issuer if its own
cardholders replaced credit card transactions with debit card transactions due to its debit rewards
program (assuming that credit card business is more profitable).
This paper seeks to analyze the effects of rewards programs on consumer payment
choices. Specifically, by using consumer survey data, we examine whether credit/debit reward
receivers use credit/debit cards relatively more often than other consumers, if so how much more
often, and which payment methods are replaced by reward card payments. We restrict our
analysis to in-store payments because the heterogeneity across payment instruments for other
4
See, for example, ATM&Debit News, August 25, 2005; December 22, 2005.
Visa and MasterCard introduced new interchange rate schemes in 2005. Interchange rates of reward credit cards
(such as Visa’s Signature Card and MasterCard’s World Card) are higher than those of non-reward credit cards by
about 0.1 to 0.8 percentage points, depending on the type of merchants and transactions.
6
This is due to the no-surcharge rule. Merchants are allowed to give discounts to cash or check users.
5
3
types of transactions, such as purchases over the Internet and bill payments, is rather
remarkable. 8
Our results suggest that (i) consumers with credit card rewards use credit cards much
more exclusively than those without credit card rewards; (ii) even among those who carry a
credit card balance, consumers with credit card rewards use a credit card more often than those
without rewards; (iii) among consumers who receive credit card rewards, those who also receive
debit card rewards tend to use debit cards more often—consumers who receive rewards only
from credit cards tend to use their credit cards more exclusively, while consumers who receive
rewards from both credit and debit cards tend to distribute their transactions more ‘equally’
between credit and debit cards; and (iv) reward card transactions seem to replace not only paperbased transactions but also non-reward card transactions.
Our analysis cannot provide evidence for causality—whether that rewards encourage
consumers to use their reward cards or that consumers who know they use their cards very often
join the rewards programs—but reveals distinctive differences in payment choice according to
whether the consumer receives rewards from which (credit, debit or both) card transactions. As
more and more transactions become electronic, understanding consumer payment choice,
especially choice between credit and debit card, becomes important. Several previous studies
have suggested that consumer debit card use is explained by behavioral motives, such as to avoid
overspending on credit cards. 9 Other studies pointed out transaction type differences, such as
transaction value, types of goods purchased, and physical environment of points of sale, affect
7
According to the 2005/2006 Study of Consumer Payment Preference, holding a reward credit card and income are
positively correlated. According to the SCF (2001), the credit card penetration rate is lower among low-income
households.
8
For example, it is difficult to make cash or check transactions for purchases over the Internet and some billers do
not accept debit cards when consumers pay bills.
9
Those studies include Ausubel (1991), Prelec and Simester (2001), and Bertaut and Haliassos (2002)
4
the choice between credit and debit card. 10 Zinman (2004) concluded that although consumer
debit usage can be partly explained by behavioral motives, it is also explained by consumer’s
cost minimization—consumers who carry a balance on their credit cards use debit in order to
reduce their interest costs on the credit card. This paper adds another possible determinant of
consumer payment choice between credit and debit cards, which is receiving rewards on the
card.
The rest of the paper is organized as follows. Section 2 describes our data set. Empirical
models are constructed in section 3. Section 4 presents results and discusses implications of the
results. Section 5 concludes the paper.
2.
Data
Our data set is the 2005/2006 Study of Consumer Payment Preference conducted by the
American Bankers Association and Dove Consulting. Data were gathered using paper and Webbased surveys sent to U.S. consumers in 2005. 3008 completed surveys were received; of them,
2350 were submitted via the Web and 658 on paper. Although the survey sample is not
nationally representative, the survey contains rich information on consumer payments, which is
usually unobservable in the nationally representative data sources. 11
First, our data set includes information on whether the consumer received credit card
rewards and debit card rewards, respectively. This allows us to examine whether credit/debit
reward receivers’ payment choice is different from non-reward receivers’.
Second, the survey asked consumer’s payment usage in detail. Those questions includes
how many times per week the consumer used each of the six payment methods at the point of
10
11
For example, see Hayashi and Klee (2003) and Klee (2006b)
For instance, Survey of Consumer Finance (SCF) conducted triennially by the Federal Reserve.
5
sale—cash, check, credit card, PIN-based debit card, signature-based debit card, and prepaid
card—and the most frequently used payment method by retail type. 12
Lastly, in the survey, consumers were asked to provide their perceptions toward each of the
six in-store payment methods. Typically, a consumer’s perceptions are not easily observed, and
therefore in a typical empirical study, they are treated as part of the consumer’s unobservable
heterogeneity. 13 But these questions allow us to observe more detailed consumer heterogeneity.
We construct our sample by excluding missing information regarding consumer
characteristics, perceptions toward in-store payment methods, card related status, such as a
balance on credit card and rewards on credit and/or on debit cards. We, then, exclude responses
that do not have a bank account and do not hold either a credit or debit card because we want to
emphasize the difference in payment choice between reward receivers and non-reward receivers,
not between card holders and non-card holders. This process leaves us 1979 responses.
Compared with the general U.S. population, our sample is relatively higher educated and higher
income (Table 1).
Table 2 shows statistics on reward receivers in our sample. About 36 percent of consumers
receive rewards via either credit card, debit card, or both. Approximately two-thirds of them (23
percent of consumers in our sample) receive rewards from credit card only; one-quarter of them
(9 percent of consumers in our sample) receive rewards from both credit and debit cards; and the
rest (4 percent of consumers in our sample) receive rewards from debit card only. The majority
of debit reward receivers receive rewards when they make signature-based debit transactions
while only 50 percent of debit reward receivers receive rewards when they make PIN-based
transactions.
12
13
See Hayashi, Sullivan, and Weiner (2003) for the difference between PIN-based and signature-based debit.
Mantel (2000) utilized consumer perceptions to estimate consumers’ bill payment choices.
6
Table 3 provides consumer characteristics and perceptions toward payment methods that
are correlated with consumer reward card holdings. Consumers with higher income and higher
educational level are more likely to hold a reward credit card. Income and educational level seem
not to affect consumers’ reward debit card holdings. Young consumers whose age is between 25
and 34 years old are more likely to hold reward debit cards. Age, however, seems to have no
effects on reward credit card holdings. Both reward credit card and reward debit card holdings
are affected by consumer ethnicity, residential region, and technology adoption behavior: Asian
and Caucasian are more likely to hold reward credit cards; People living in New England are
more likely to hold reward credit cards and people living in Mid-Atlantic region are more likely
to hold reward debit cards; Consumers who use Internet at work and online banking are more
likely to receive rewards from both credit and debit cards. Consumer perception toward payment
cards and reward card holdings are correlated. Naturally, consumers who have positive
perceptions toward a credit card tend to have reward credit cards and consumers who have
positive perceptions toward a debit card tend to have reward debit cards. Interestingly,
consumers who have relatively negative perceptions toward debit cards tend to receive credit
rewards.
The next two figures indicate that credit card rewards and consumer payment choice have a
strong correlation. Not only credit card rewards, but also debit card rewards and credit card
balance seem to be correlated with consumer payment choice.
Figure 1(a) shows consumer distribution in terms of the share of credit card transactions in
the total in-store transactions. Consumers are divided into eight groups according to whether they
have a credit card balance, whether they receive debit card rewards, and whether they receive
credit card rewards. All groups but one have a similar distribution pattern: the percent frequency
7
(consumer density) declines as the share of credit card transactions increases. But one group—
consumers who do not carry a credit card balance, do not receive debit card rewards but receive
credit card rewards—has an almost uniform distribution. Clearly, consumers in this group use a
credit card more exclusively. Figure 1(b) shows consumer distribution in terms of the share of
debit card transactions. Similar to figure 1(a), only one group of consumers has a distinctive
distribution. That group consists of consumers who do not carry a credit card balance, do not
receive debit card rewards, and receive credit card rewards. The majority of consumers in the
group have a debit card share that is less than 50 percent. In other groups of consumers, on the
other hand, they are distributed more evenly, meaning that many consumers in these groups use a
debit card more often. Figure 1(c) shows consumer distribution in terms of the share of paperbased transactions. In contrast with previous two figures, in this figure there is no group of
consumers that reveals a distinctive distribution pattern.
Figure 2 presents the share of consumers who choose a particular payment instrument as
their most frequently used payment method at grocery stores. Similar to figure 1, consumers are
grouped into eight groups, according to their credit card balance, debit card rewards, and credit
card rewards status. The top panel shows credit card, the middle is debit card, and the bottom is
paper based payments as the most frequently used payment method. From the credit card panel,
we see that one group exclusively choose the credit card as their most frequently used payment
method at grocery stores. The group is consumers who do not carry a credit card balance, do not
receive debit card rewards, and receive credit card rewards. In the other groups, at most 25
percent of consumers choose the credit card as their most frequently used method. As can be
guessed, most of consumers in the group who do not carry a credit card balance, do not receive
debit card rewards, and receive credit card rewards, do not choose debit card as their most
8
frequently used payment method. From the panel of paper-based payment method, non-receivers
of credit card rewards tend to choose paper-based payments as their most frequently used
payment method at grocery stores.
3.
Model
Consumer payment choice is influenced by various factors. Previous research has
highlighted three important sets of factors: consumer characteristics, payment method attributes
and transaction characteristics.
Consumer characteristics, such as age, income, and educational level, have shown to be
correlated with use of payment methods in previous literature. Adoption of other technologies
also influences consumer payment choice. These factors could proxy for preferences (checks are
preferred more by women than by men), for availability of payment methods (consumers with
higher income are more likely to use credit cards than those with lower income), and for
familiarity with new payment technologies (people who use new technologies are more likely to
use debit cards and to make transactions over the Internet).
Payment method attributes may also be important determinants when consumers choose a
payment method. Some payment instruments have distinctive attributes. For example, some
payment cards offer rewards to their users. Cash gives consumers anonymity, while the other
payment methods don’t. Credit cards provide liquidity at least until the next billing date. Other
attributes, such as transaction speed, safety, and ease of use, vary by payment method.
Transaction characteristics, such as value of the transaction and physical environment,
likely influence consumer payment choice. For example, consumers tend to pay with cash for a
smaller value transaction, while they tend to use a credit card or a check for a larger value
transaction. The effects of physical environment on the use of payment may partly be supply side
9
effects. Some types of retail stores may not accept all or some types of payment cards. Only cash
may be accepted when consumers use public transportation systems, such as a city bus or the
subway. Even when merchants accept all payment methods, some payments may be less
convenient to use than the others. For example, at a restaurant consumers cannot make a PINbased debit payment at their seat unless the restaurant carries a portable PIN pad. At a gas
station, consumers may not need to go to the cashier if they pay with cards, while they need to do
so if they pay with cash or checks.
Our data set does not contain information on transaction characteristics. However, we can
observe each consumer’s most frequently used payment method by retail type. At a certain type
of retail store, variation of transaction characteristics may be limited.
Information on payment method attributes per se is not included in our data set, either.
However, we can observe each consumer’s perceptions toward each in-store payment method.
Those perceptions include speed, safety, ease of use, comfort, convenience, help budget, etc. We
see them as payment method attributes evaluated by each consumer.
Utility to consumer i from using payment method j when making a transaction at retail type
h is defined as follows:
U ijh = X i β jh + Z ij γ h + C ij δ jh + ε ijh ,
(1)
where X i is a vector of consumer characteristics, Z ij is a vector of attributes of payment method
evaluated by the consumer and Cij is a vector of card-related dummies. β , γ , and δ are vectors
of parameters that weight these factors in the consumer’s utility function. ε ijh captures
unobserved factors and is assumed to be i.i.d. extreme value.
We consider four card-related dummies: rewards on credit, on PIN-based debit, and on
signature-based debit, and a balance on the credit card. If a consumer receives rewards on a
10
credit card, her utility from using the credit card will be higher than otherwise. Her utility from
using another payment method will not be affected by whether she receives credit rewards or not.
Similarly, PIN-based debit rewards and signature-based debit rewards affect her utility from
using PIN debit and from using signature debit, respectively. If the consumer carries a positive
balance on a credit card, her utility from using the credit card will be lower because it costs her.
We allow coefficients on reward dummies to vary by payment method, because credit card
rewards are typically more generous than debit card rewards.
Because consumers who chose prepaid card as their most frequently used payment method
are negligible, we excluded such consumers from our sample. Thus, consumers have five
payment options: credit card, PIN-based debit card, signature-based debit card, check, and cash.
In addition to the above discrete choice models, we also estimate each consumer’s share of
transactions that used a particular payment method in the consumer’s total in-store transactions,
which is derived from the weekly frequency question. Although the discrete choice models allow
us to measure the difference between reward receivers and non reward receivers in terms of the
likelihood of choosing a certain payment method as their most frequently used payment method,
they cannot be used to measure the intensity of a certain payment method usage. Estimating the
share, on the other hand, allows us to measure how much more or less reward receivers use a
particular payment method relative to non-reward receivers, if each consumer’s consumption
basket is alike. We use a series of least square estimations to estimate the share.
4.
Results
4.1
Estimation of the most frequently used payment method by retail type
Table 4 reports the estimation results for the most frequently used payment method at
grocery stores. We estimate four model specifications. The first specification includes only
11
consumer characteristics in order to offer a point of comparison to previous studies and to our
other specifications. The second includes card related variables, such as credit and debit reward
dummies and a credit card balance dummy, in addition to the variables in the first specification.
The third specification includes perception variables instead of card related variables. Finally, the
fourth specification includes both card related and perception variables as well as consumer
characteristics.
Since our model is a multinomial logit model, coefficients are not easily interpreted.
Nevertheless, in the first specification coefficients on gender, age, income, educational level,
race, and technology adoption variables are statistically significantly different from zero. This
implies that these consumer characteristics influence consumer payment choice.
By adding three rewards dummies and a credit card balance dummy, the log-likelihood
increases from -2689 to -2571. In this specification, coefficients on all reward dummies and
credit balance dummy are statistically significantly different from zero. Signs of these
coefficients are as expected: Rewards increase the probability that consumers choose to pay with
the card with rewards and consumers without credit card balance are more likely to choose credit
card.
By adding perception variables to consumer characteristics, the log-likelihood is improved
significantly (from -2689 to -1689). This implies that consumer perceptions are important to
predict the probability of choosing the most frequently used payment method at grocery stores.
In the full specification, the effect of payment card rewards and credit balance is
diminished. The coefficients on credit and signature-based debit reward dummies and credit
balance dummy are still significant, but that on PIN-based debit reward dummy becomes
insignificant. This may suggest that the PIN debit reward dummy acts as a proxy for consumer’s
12
better perception toward PIN-based debit. Therefore, after controlling for consumer perception, it
has little effect on consumer payment choice. In contrast, credit rewards, signature debit rewards,
and a credit card balance have significant effects on the consumer use of payment, even after
controlling for consumer perception.
We also estimate models of the most frequently used payment method at other types of
stores—department stores, discount stores, drug stores, and fast food restaurants. 14 In the full
specification, only credit rewards have a significant effect on consumer payment choice at all
types of stores. Signature debit rewards and a credit card balance have a significant effect at
some types of stores, while PIN debit rewards do not have a significant effect at any types of
stores.
We now quantify the effects of payment card rewards on consumer payment choice. To do
so, we construct predicted probability of using different payment methods according to whether
the consumer receives rewards on credit card and/or on signature debit card. The predicted
probability varies by consumer and by retail type. Figure 3 shows these probabilities for middleaged Caucasian male college graduates with income $70,000 who do not carry a credit card
balance at five different retail types. 15 We consider four cases: 1) rewards on neither credit nor
debit card; 2) rewards on credit card only; 3) rewards on (signature-based) debit card only; and
4) rewards on both credit and debit cards.
Before analyzing four cases, it may be worth noting that the same consumer’s probabilities
of choosing a payment method significantly vary by retail type. The consumer tends to use a
credit card more exclusively at department stores, while at fast food restaurants he tends to use
cash more exclusively. For the other three types of stores, his probability of choosing a payment
14
Specification 4 results are shown in appendix.
13
method from a credit card, a debit card, and a paper-based payment (cash or check) is relatively
more evenly distributed.
Compared with the case where the consumer does not receive rewards at all (case 1), when
he receives rewards on credit card only (case 2) his probability of choosing a credit card is
greater at any of the five types of retail stores. As a result, not only his probability of choosing a
paper-based payment but also the probability of choosing a debit card become smaller. The
effects of credit card rewards vary by retail type. The reward credit card holder’s probability of
choosing a credit card is about 25 percentage points greater at grocery stores, approximately 15
percentage points greater at department stores, discount stores and drug stores, and only 8
percentage points greater at fast food restaurants. The decline in probability of choosing a debit
card surpasses the decline in probability of choosing a paper-based payment at grocery stores and
department stores.
Compared with case 1, when the consumer receives rewards on debit card only (case 3), his
probability of choosing a debit card becomes greater. Both probabilities of choosing a credit card
and of choosing a paper-based payment become smaller. Similar to the effects of credit card
rewards, the effects of debit card rewards vary by retail type. The reward debit card holder’s
probability of choosing a debit card is about 10 percentage points greater at grocery stores and
department stores, 5 to 7 percentage points greater at discount stores and drug stores, and less
than 1 percentage point greater at fast food restaurants. At all types of stores but department
stores, the decrease in probability of choosing a paper-based payment is greater than the decrease
in probability of choosing a credit card.
15
As these consumers’ other consumer characteristics and perceptions, we take the average of male Caucasian
college graduate consumers.
14
As we have seen, the effects of credit card rewards seem to be larger than the effects of
debit card rewards. For example, at grocery stores, the consumer’s probability of choosing a
credit card increases more than 20 percentage points, while his probability of choosing a debit
card increases about 10 percentage points. This may imply that consumers will react to rewards
on credit card more enthusiastically than to rewards on debit card.
Compared with case 1, when the consumer receives rewards on both credit and debit cards
his probability of choosing a credit card is always greater, but his probability of choosing a debit
card is not always greater. Rather, his probability of using a debit card is smaller at grocery
stores, department stores and discount stores.
We also compare the probabilities with other consumer characteristics. Consumers with
different characteristics, such as consumers with a credit card balance, female consumers, and
consumers with less educational level, show similar patterns of probability differences as we
compare four cases. Although the pattern is similar, the reaction to rewards by those consumers
is somewhat more moderate.
Our findings suggest four things. First, the effects of rewards on consumer payment choice
may vary by retail type. Second, reward card transactions—either credit or debit card rewards—
may replace not only paper-based transactions but also non-reward card transactions. Third, the
effects of credit card rewards may surpass the effects of debit card rewards. Fourth, even among
consumers who carry a positive credit card balance, consumers who receive rewards on credit
card may be more likely to use credit card than those who do not receive rewards on credit card.
15
4.2
Share estimation
Table 5 presents the results of the share estimation. 16 Similar to the multinomial logit
models, four model specifications are used. For all of our specifications, the baseline race
category is Caucasian, the baseline age category is 35-44 years old, the baseline education
category is college, the baseline income category is less than $40,000 and the baseline census
division is East South Central.
Our results from the first specification are basically aligned with findings in the previous
studies. Female consumers tend to use debit cards more frequently than male consumers. African
Americans are heavy users of paper-based payments, while Asian Americans use credit cards
more exclusively. Younger consumers use both credit and debit cards more often than older
consumers. Interestingly, both consumer groups whose education level is high school or less and
higher than college tend to use debit cards less often than the consumer group whose education
level is college, while consumers with higher education tend to use credit card more often. Credit
card share does not vary by region very much; however debit card share and paper-based
payments share vary by region. Technology adoption variables significantly affect debit card
share (positively) and paper-based payments share (negatively). Although demographic,
financial, regional, and technology adoption characteristics significantly affect consumer
payment choice, the adjusted R-square of this specification is at most 0.1.
From the second specification, we see that credit and debit rewards dummies significantly
affect the shares of credit card, debit card, and paper-based transactions. The signs of coefficients
on credit card rewards and on debit card rewards are as expected: the coefficients on credit card
rewards is positive for the credit card share, but it is negative for the share of debit card and of
16
All of the results shown in table 5 used OLS models, but we will modify the models to more sophisticated ones,
such as GLS and with instrumental variables.
16
paper-based transactions; the coefficient on debit card reward is positive for the debit card share,
but it is negative for the share of credit card and of paper-based transactions. The magnitude of
coefficients on both rewards dummies is greater than that of most other coefficients on
demographic, financial, regional, and technology adoption variables. The credit card balance
dummy is statistically significant for the credit and debit card share estimations. Consumers who
have a positive balance on credit cards use debit cards more often and consumers who do not
carry a balance use credit cards more often. Compared with the first specification, adding three
card-related dummies significantly improves the adjusted R-squares for the credit card and debit
card share estimations.
The results from the third specification suggest that consumer perceptions toward payment
instruments significantly predict consumer payment usage. The adjusted R-square of the third
specification is at least three times as high as that of the first specification. Naturally, consumers
who have better perceptions toward a certain payment method use the payment method more
often. For example, consumers who feel comfortable with a credit card transaction tend to use a
credit card more often and to use a debit card or a paper-based payment less often.
By adding the three card-related variables to the third specification, the adjusted R-square
is not increased very much. Nevertheless, the coefficients on the three card-related dummies are
statistically significant for both credit and debit card shares in the fourth specification. Two of
the three dummies (credit card balance and credit card rewards) significantly affect the share of
paper-based payments. This implies that whether a consumer receives rewards from credit card
transactions and/or from debit card transactions is important to predict the consumer’s payment
usage.
17
According to our specification 4 results, receiving rewards on credit card represents a 9.0
percentage point increase in credit card share, a 5.6 percentage point decrease in debit card share,
and a 4.1 percentage point decrease in paper-based payment share. Although either the increase
in credit card share is underestimated or the decrease in debit card share and/or in paper-based
payment share is overestimated, the results suggest that reward credit card transactions likely
replace both debit transactions and paper-based transactions. Receiving rewards on debit card
represents a 6.9 percentage point increase in debit card share, a 6.2 percentage point decrease in
credit card share, and a 0.9 percentage point decrease in paper-based payment share. This may
imply that majority of reward debit card transactions replace credit card transactions. The effects
of credit card rewards seem to be larger than the effects of debit card rewards.
5.
Conclusion
This paper shows that consumer payment choice is correlated with rewards on payment
cards. Our results suggest that (i) consumers with credit card rewards use credit cards more
exclusively than those without credit card rewards; (ii) even among those who carry a credit card
balance, consumers with credit card rewards use a credit card more often than those without
rewards; (iii) among consumers who receive credit card rewards, those who receive rewards on
credit cards as well as on debit cards tend to use debit cards more often than those who receive
rewards only on credit cards; and (iv) reward card transactions seem to replace not only paperbased transactions but also non-reward card transactions.
We also show that the effects of payment card rewards vary by consumer and retail type.
Because we cannot observe how many in-store transactions take place at each type of retail
stores it is difficult to examine how payment card rewards affect overall consumer in-store
payment choice. Moreover, consumer payments are not just in-store payments but also payments
18
over Internet, such as electronic bill payments and e-commerce payments. It is very likely that
the payment card rewards affect consumer payment choice differently according to whether the
transactions are in-store transactions, electronic bill payments, or e-commerce transactions. More
comprehensive analysis is needed to understand how payment rewards affect overall consumer
payment choice.
19
References
Ausbel, Lawrence M. 1991. “The Failure of Competition in the Credit Card Market,” American
Economic Review, 81(1): 50-81.
Bertaut, Carol C. and Michael Haliassos. 2002. “Debt Revolvers for Self-Control,” University of
Cyprus Working Papers in Economics 0209.
Borzekowski, Ron and Elizabeth Kiser. 2006. “The Choice an the Checkout: Quantifying
Demand Across Payment Instruments” Federal reserve Board Finance and Economics
Discussion Series, 2006-17.
Dove Consulting and the American Bankers Association. 2005. The 2005/2006 Study of
Consumer Payment Preferences.
Hayashi, Fumiko and Elizabeth Klee. 2003. “Technology Adoption and Consumer Payments:
Evidence from Survey Data,” Review of Network Economics, 2(2): 175-190.
Hayashi, Fumiko, Richard Sullivan, Stuart E. Weiner. 2003. A Guide to the ATM and Debit Card
Industry, Federal Reserve Bank of Kansas City.
Hirschman, Elizabeth. 1982. “Consumer Payment Systems: The Relationship of Attribute
Structure to Preference and Usage” Journal of Business, 55(4): 531-545.
Jonker Nicole. 2005. “Payment Instruments as Perceived by Consumers—a Public Survey” De
Nederlandsche Bank Working Paper No. 053/2005.
Klee, Elizabeth. 2006a “Families’ Use of Payment Instruments During a Decade of Change in
the US Payment System” Federal reserve Board Finance and Economics Discussion
Series, 2006-1.
Klee, Elizabeth. 2006b “Paper or Plastic? The Effect of Time on the Use of Check and Debit
Cards at Grocery Stores” Federal reserve Board Finance and Economics Discussion
Series, 2006-2.
Mantel, Brian. 2000. “Why Do Consumers Pay Bills Electronically? An Empirical Analysis,”
Federal Reserve Bank of Chicago Economic Perspectives, 32-47.
Prelac, Drazen and Duncan Simester. 2001. “Always Leave Home without It: A Further
Investigation of the Credit Card Effect on Willingness to Pay,” Marketing Letters, 12 (1):
5-12.
Reda, Suzan. 2003. “2003 Consumer Credit Survey,” STORES Magazine, November.
Zinman, Jonathan. 2004. “Why Use Debit Instead of Credit? Consumer Choice in Trillion-Dollar
Market,” Federal Reserve Bank of New York Staff Report no.191.
20
Table 1: Summary statistics
Our sample
Census
.491
.514
.117
.067
.703
.070
.043
.120
.040
.700
.120
.020
.269
.253
.174
.208
.096
.315
.200
.189
.132
.165
.011
.536
.311
.142
.160
.510
.250
.080
.352
.240
.282
.126
.463
.178
.209
.151
.050
.118
.209
.050
.106
.168
.072
.070
.158
.051
.142
.191
.061
.105
.160
.069
.058
.163
Demographic
Female
Race
African American
Asian
Caucasian
Hispanic
Other
Age
18-34
35-44
45-54
55-64
65 and over
Education
Less than high school
High school
College
Some graduate school
Financial (Income)
$0 - $40,000
$40,000 - $59,999
$60,000 - $99,999
$100,000 and over
Census division
New England
Mid Atlantic
South Atlantic
ES Central
EN Central
WS Central
WN Central
Mountain
Pacific
21
Table 2: Reward card holders
Rewards card holders
Reward credit
Reward debit
Reward PIN debit
Reward signature debit
Reward credit & debit
Reward credit & PIN debit
Reward credit & signature debit
Sample size
721
Percent of
sample
36.43
Percent of
reward holders
100
634
269
131
242
32.03
13.59
6.62
12.28
87.93
37.31
18.17
33.56
182
82
167
9.20
4.14
8.44
25.24
11.37
23.16
Table 3: Characteristics of reward card holders
Credit card reward
Asian
Caucasian
Income>$60,000
Education>college
Living in New England
Technology users
With credit card perception
- comfortable
- fast
- convenient
- easy to use
- preferred by stores
- safe
- spend within my means
- for small amounts
- control over money
- easy to get refund
PIN debit card reward
Asian
25<=Age<=34
Living in Mid-Atlantic
Users of Internet at work
With PIN and signature debit
card perception
- comfortable
- fast
22
Signature debit reward
25<=Age<=34
Living in Mid-Atlantic
Users of Internet at work
Users of online banking
With PIN and signature debit
card perception
- comfortable
- fast
With signature debit card
perception
- convenient
- easy to use
- preferred by stores
- safe
- spend within my means
- for small amounts
- control over money
- easy to get refund
- money is taken from
account right away
Figure 1(a): Consumers’ distribution in terms of their share of credit transactions:
grouped by credit card balance, debit rewards, and credit rewards
Credit
ccwob=0 dcwr=0 ccwr=0
% freq
Credit
ccwob=0 dcwr=0 ccwr=1
% freq
60
60
50
50
40
40
30
30
20
20
10
10
0
0
0-
0.1-
0.2- 0.3-
0.4-
0.5-
0.6-
0.7- 0.8-
0-
0.9-
0.1-
0.2-
0.3-
0.4- 0.5-
Share
Credit
ccwob=0 dcwr=1 ccwr=0
% freq
0.6-
0.7-
0.8-
0.9-
0.7-
0.8-
0.9-
0.7-
0.8-
0.9-
Share
Credit
ccwob=0 dcwr=1 ccwr=1
% freq
60
60
50
50
40
40
30
30
20
20
10
10
0
0
0-
0.1-
0.2-
0.3-
0.4- 0.5-
0.6-
0.7-
0.8-
0.9-
0-
0.1-
0.2-
0.3-
0.4-
Share
Credit
ccwob=1 dcwr=0 ccwr=0
% freq
0.5-
0.6-
Share
Credit
ccwob=1 dcwr=0 ccwr=1
% freq
60
60
50
50
40
40
30
30
20
20
10
10
0
0
0-
0.1-
0.2-
0.3-
0.4-
0.5-
0.6-
0.7-
0.8-
0.9-
0-
0.1-
0.2-
0.3-
0.4-
Share
Credit
ccwob=1 dcwr=1 ccwr=0
% freq
50
40
30
20
10
0
0.1-
0.2-
0.3-
0.4-
0.5-
0.6-
0.6-
Credit
ccwob=1 dcwr=1 ccwr=1
% freq
60
0-
0.5-
Share
0.7-
0.8-
0.9-
90
80
60
70
50
60
5040
4030
3020
2010
10
0
0
East
West
North
0-
0.1-
0.2-
0.3-
0.4-
0.5-
0.6-
1st Qtr 2nd Qtr 3rd Qtr 4th Qtr
Share
Share
Notes: ccwob=1, if consumers do not carry a credit card balance; ccwob=0, otherwise.
dcwr=1, if consumers receive debit card rewards; dcwr=0, otherwise.
ccwr=1, if consumers receive credit card rewards; ccwr=0, otherwise.
23
0.7-
0.8-
Figure 1(b): Consumers’ distribution in terms of their share of debit transactions:
grouped by credit card balance, debit rewards, and credit rewards
Debit
ccwob=0 dcwr=0 ccwr=0
% freq
Debit
ccwob=0 dcwr=0 ccwr=1
% freq
60
60
50
50
40
40
30
30
20
20
10
10
0
0
0-
0.1-
0.2-
0.3-
0.4- 0.5-
0.6-
0.7-
0.8-
0.9-
0-
0.1-
0.2-
0.3-
Share
0.5-
0.6-
0.7-
0.8-
0.9-
0.7-
0.8-
0.9-
0.7-
0.8-
0.9-
0.7-
0.8-
0.9-
Share
Debit
ccwob=0 dcwr=1 ccwr=0
% freq
0.4-
Debit
ccwob=0 dcwr=1 ccwr=1
% freq
60
60
50
50
40
40
30
30
20
20
10
10
0
0
0-
0.1-
0.2-
0.3-
0.4-
0.5-
0.6-
0.7-
0.8-
0.9-
0-
0.1-
0.2-
0.3-
Share
0.5-
0.6-
Share
Debit
ccwob=1 dcwr=0 ccwr=0
% freq
0.4-
Debit
ccwob=1 dcwr=0 ccwr=1
% freq
60
60
50
50
40
40
30
30
20
20
10
10
0
0
0-
0.1-
0.2-
0.3-
0.4-
0.5-
0.6-
0.7-
0.8-
0.9-
0-
0.1-
0.2-
0.3-
Share
0.5-
0.6-
Share
Debit
ccwob=1 dcwr=1 ccwr=0
% freq
0.4-
Debit
ccwob=1 dcwr=1 ccwr=1
% freq
60
60
50
50
40
40
30
30
20
20
10
10
0
0
0-
0.1-
0.2-
0.3-
0.4-
0.5-
0.6-
0.7-
0.8-
0.9-
0-
Share
0.1-
0.2-
0.3-
0.4-
0.5-
Share
Notes: ccwob=1, if consumers do not carry a credit card balance; ccwob=0, otherwise.
dcwr=1, if consumers receive debit card rewards; dcwr=0, otherwise.
ccwr=1, if consumers receive credit card rewards; ccwr=0, otherwise.
24
0.6-
Figure 1(c): Consumers’ distribution in terms of their share of paper-based transactions:
grouped by credit card balance, debit rewards, and credit rewards
Paper-based
ccwob=0 dcwr=0 ccwr=0
% freq
Paper-based
ccwob=0 dcwr=0 ccwr=1
% freq
60
60
50
50
40
40
30
30
20
20
10
10
0
0
0-
0.1-
0.2-
0.3-
0.4- 0.5-
0.6-
0.7-
0.8-
0.9-
0-
0.1-
0.2-
0.3-
Share
0.5-
0.6-
0.7-
0.8-
0.9-
0.7-
0.8-
0.9-
0.7-
0.8-
0.9-
0.7-
0.8-
0.9-
Share
Paper-based
ccwob=0 dcwr=1 ccwr=0
% freq
0.4-
Paper-based
ccwob=0 dcwr=1 ccwr=1
% freq
60
60
50
50
40
40
30
30
20
20
10
10
0
0
0-
0.1-
0.2-
0.3-
0.4-
0.5-
0.6-
0.7-
0.8-
0.9-
0-
0.1-
0.2-
0.3-
Share
0.5-
0.6-
Share
% freq
Paper-based
ccwob=1 dcwr=0 ccwr=0
% freq
0.4-
60
60
50
50
40
40
30
30
20
20
10
10
0
Paper-based
ccwob=1 dcwr=0 ccwr=1
0
0-
0.1-
0.2-
0.3-
0.4-
0.5-
0.6-
0.7-
0.8-
0.9-
0-
0.1-
0.2-
0.3-
Share
% freq
0.4-
0.5-
0.6-
Share
Paper-based
ccwob=1 dcwr=1 ccwr=0
Paper-based
ccwob=1 dcwr=1 ccwr=1
% freq
60
60
50
50
40
40
30
30
20
20
10
10
0
0
0-
0.1-
0.2-
0.3-
0.4-
0.5-
0.6-
0.7-
0.8-
0.9-
0-
Share
0.1-
0.2-
0.3-
0.4-
0.5-
Share
Notes: ccwob=1, if consumers do not carry a credit card balance; ccwob=0, otherwise.
dcwr=1, if consumers receive debit card rewards; dcwr=0, otherwise.
ccwr=1, if consumers receive credit card rewards; ccwr=0, otherwise.
25
0.6-
Figure 2: Share of the payment method as the most frequently used method at the grocery
stores by credit card balance, debit rewards, and credit rewards
Credit Card
ccwob=0, dcwr=0, ccwr=0
ccwob=0, dcwr=0, ccwr=1
ccwob=0, dcwr=1, ccwr=0
ccwob=0, dcwr=1, ccwr=1
ccwob=1, dcwr=0, ccwr=0
ccwob=1, dcwr=0, ccwr=1
ccwob=1, dcwr=1, ccwr=0
ccwob=1, dcwr=1, ccwr=1
0%
10%
20%
30%
40%
50%
60%
70%
50%
60%
70%
60%
70%
Credit
Debit Card
ccwob=0, dcwr=0, ccwr=0
ccwob=0, dcwr=0, ccwr=1
ccwob=0, dcwr=1, ccwr=0
ccwob=0, dcwr=1, ccwr=1
ccwob=1, dcwr=0, ccwr=0
ccwob=1, dcwr=0, ccwr=1
ccwob=1, dcwr=1, ccwr=0
ccwob=1, dcwr=1, ccwr=1
0%
10%
20%
30%
PIN
40%
Signature
Paper-based
ccwob=0, dcwr=0, ccwr=0
ccwob=0, dcwr=0, ccwr=1
ccwob=0, dcwr=1, ccwr=0
ccwob=0, dcwr=1, ccwr=1
ccwob=1, dcwr=0, ccwr=0
ccwob=1, dcwr=0, ccwr=1
ccwob=1, dcwr=1, ccwr=0
ccwob=1, dcwr=1, ccwr=1
0%
10%
20%
30%
Cas h
40%
50%
Check
Notes: ccwob=1, if consumers do not carry a credit card balance; ccwob=0, otherwise.
dcwr=1, if consumers receive debit card rewards; dcwr=0, otherwise.
ccwr=1, if consumers receive credit card rewards; ccwr=0, otherwise.
26
Figure 3: Predicted probability of choosing payment method by retail type
Departm ent Stores
Grocery Stores
80%
80%
70%
70%
60%
60%
50%
50%
40%
40%
30%
30%
20%
20%
10%
10%
0%
0%
non
cc only
cc
dc
dc only
cc+dc
non
cc only
cc
paper
dc
dc only
cc+dc
paper
Discount Stores
Fast Food Restaurants
80%
80%
70%
70%
60%
60%
50%
50%
40%
40%
30%
30%
20%
20%
10%
10%
0%
0%
non
cc only
cc
dc
dc only
non
cc+dc
paper
cc
Drug Stores
80%
70%
60%
50%
40%
30%
20%
10%
0%
non
cc only
cc
dc
dc only
cc only
cc+dc
paper
27
dc
dc only
paper
cc+dc
Table 4: Multinomial logit model for payment choice at grocery stores
Specification
1
Credit
PIN debit
2
Signature debit
Check
Credit
PIN debit
Signature debit
Check
Card related
.800**
1.63**
No balance on credit
Rewards on credit
Reward on PIN
Reward on Sig
Demographic
Female
Race
Asian
Other
Age
Age^2
Education
Education^2
Income
Income^2
Technology
Direct deposit
Online banking
Intercept
.147
.145
.563**
.202
1.63**
.174
.365**
.165
.582**
.137
.619**
.174
1.09**
.176
.425**
.173
.588**
.138
.613**
.179
1.09**
.176
.238
-1.15**
-.111**
.001**
.194
.051
.159**
-.005*
.280
.217
.033
.000
.543
.077
.058
.003
-1.40**
-.627**
-.011
.000
.364
-.023
.111**
-.005*
.323
.158
.029
.000
.425
.063
.050
.003
-1.71**
-.675**
.008
.000
1.34**
-.161*
.070*
-.003
.480
.201
.040
.000
.610
.089
.064
.003
-.697*
-.999**
.048*
-.000
-.105
.035
.235**
-.013**
.435
.231
.039
.000
.521
.076
.069
.004
.376*
-.961**
-.095**
.001**
.208
.035
.116*
-.005*
.286
.231
.034
.000
.568
.081
.060
.003
-1.47**
-.647**
-.012
.000
.369
-.024
.114**
-.006**
.320
.158
.029
.000
.410
.061
.050
.003
-1.82**
-.730**
.015
-.000
1.27*
-.150*
.072*
-.004
.511
.208
.041
.000
.664
.096
.068
.004
-.731*
-1.01**
.046*
-.000
-.075
.029
.238**
-.013**
.459
.232
.041
.000
.546
.081
.069
.004
.584**
.672**
.193
.167
.556**
.825**
.155
.139
.956**
.879**
.219
.181
.200*
.012
.195
.176
.584**
.672**
.193
.167
.556**
.825**
.155
.139
.939**
.772**
.224
.185
.198*
-.002
.202
.181
-.360
1.18
-1.10*
.957
-4.02**
1.34
-3.23*
1.27
-1.42*
1.24
-1.12*
.911
-4.29**
1.42
-3.22**
1.29
Perceptions
Comfortable
Fast
Convenient
Easy to use
Preferred by stores
Safe
Taken right away
Help me budget/spend
For small amounts
Control over money
Easy to get refund
Log-likelihood
-2688.99
N
Standard errors are in parentheses.
-2570.88
1915
**
1915
*
t-value is greater than 2. t-value is greater than 1
28
Table 4: Multinomial logit model for payment choice at grocery stores
Specification
3
Credit
PIN debit
4
Signature debit
Check
Credit
PIN debit
Signature debit
Check
Card related
.516**
1.01**
No balance on credit
Rewards on credit
Reward on PIN
Reward on Sig
Demographic
Female
Race
Asian
Other
Age
Age^2
Education
Education^2
Income
Income^2
Technology
Direct deposit
Online banking
Intercept
.175
.174
.150
.237
1.19**
.208
.574**
.193
.550**
.165
.565**
.208
1.10**
.210
.579**
.196
.523**
.166
.537**
.214
1.07**
.211
-.195
-1.15**
-.104**
.001**
.506
-.019
.100*
-.003
.316
.257
.039
.000
.639
.090
.066
.003
-1.58**
-.749**
-.042*
.000
.573
-.036
.061*
-.002
.381
.186
.037
.000
.561
.082
.061
.004
-1.43**
-.863**
-.007
-.000
1.09*
-.105
.026
-.001
.569
.235
.049
.001
.783
.113
.083
.005
-.520
-.886**
.012
-.000
.086
.007
.255**
-.013**
.544
.265
.053
.001
.633
.094
.083
.005
-.078
-.956**
-.098**
.001*
.457
-.021
.066
-.003
.320
.269
.040
.000
.661
.093
.069
.004
-1.59**
-.736**
-.045*
.000
.475
-.023
.057
-.002
.379
.187
.037
.000
.563
.083
.062
.004
-1.46**
-.826**
-.001
-.000
1.02*
-.098
.021
-.001
.586
.239
.051
.001
.794
.115
.087
.005
-.555
-.888**
.009
-.000
.027
.015
.255**
-.013**
.558
.265
.053
.001
.636
.095
.083
.005
.613**
.479**
.231
.196
.499**
.432**
.188
.169
.826**
.589**
.264
.217
.200*
.012
.195
.176
.617**
.451**
.235
.201
.499**
.432**
.188
.169
.812**
.507**
.269
.222
.196
.141
.248
.223
1.18
1.45
1.24
1.26
-1.52
1.72
.027
1.57
.695
1.50
1.41
1.27
-1.73
1.74
.150
1.57
Perceptions
Comfortable
Fast
Convenient
Easy to use
Preferred by stores
Safe
Taken right away
Help me budget/spend
For small amounts
Control over money
Easy to get refund
.696**
.329**
.813**
.642**
.274**
.148*
-.081
.316**
.227**
.628**
.179**
Log-likelihood
-1688.74
N
Standard errors are in parentheses.
.670**
.323**
.809**
.650**
.275**
.132*
-.041
.329**
.239**
.568**
.154*
.059
.051
.127
.130
.088
.078
.093
.089
.097
.093
.084
-1650.80
1915
**
.060
.052
.129
.131
.090
.080
.093
.090
.098
.095
.085
1915
*
t-value is greater than 2. t-value is greater than 1
29
Table 5: Share estimation results
Credit card
Specification
1
2
Debit card
Specification
3
4
1
2
3
4
Card related
.030**
-.097**
.158**
No balance on credit
Rewards on debit
Rewards on credit
(.009)
(.014)
(.010)
.027**
-.059**
.083**
(.008)
(.012)
(.009)
-.036**
.152**
-.135**
(.011)
(.017)
(.013)
-.009*
.067**
-.053**
(.009)
(.013)
(.011)
Perceptions
Comfortable
Credit card
Debit card
Cash or checks
Fast
Credit card
Debit card
Cash or checks
Convenient
Credit card
Debit card
Cash or checks
Help me budget/spend
Credit card
Debit card
Cash or checks
Control over money
Credit card
Debit card
Cash or checks
Demographic
Female
Race
African American
Asian
Hispanic
Other
Age
18-34
45-54
55-64
65 and over
.029**
-.024**
-.030**
(.003)
(.004)
(.006)
.027**
-.020**
-.031**
(.003)
(.004)
(.006)
-.015**
.046**
-.010*
(.004)
(.005)
(.007)
-.013**
.042**
-.009*
(.004)
(.005)
(.007)
.012**
-.006*
-.009*
(.004)
(.004)
(.005)
.013**
-.007*
-.010**
(.004)
(.004)
(.005)
.003
.017**
-.002
(.004)
(.005)
(.006)
.002
.017**
-.003
(.004)
(.005)
(.005)
.059**
-.049**
-.010
(.010)
(.011)
(.012)
.048**
-.048**
-.007
(.009)
(.011)
(.012)
-.053**
.090**
-.062**
(.011)
(.012)
(.014)
-.044**
.091**
-.062**
(.011)
(.012)
(.013)
.053**
.003
-.010
(.012)
(.010)
(.010)
.051**
.003
-.013*
(.012)
(.010)
(.010)
-.039**
.040**
-.024**
(.013)
(.011)
(.012)
-.037**
.039**
-.021*
(.013)
(.011)
(.012)
.076**
-.032**
-.014*
(.011)
(.010)
(.011)
.063**
-.026**
-.008
(.011)
(.010)
(.011)
-.053**
.058**
-.021*
(.012)
(.012)
(.012)
-.047**
.055**
-.024**
(.012)
(.012)
(.012)
-.014*
(.009)
-.011*
(.009)
.005
(.008)
.004
(.008)
.045**
(.012)
.042**
(.011)
.017*
(.009)
.018**
(.009)
-.046**
.130**
.011
-.030*
(.015)
(.020)
(.019)
(.024)
-.030**
.128**
.029*
-.023*
(.014)
(.019)
(.018)
(.022)
-.025*
.048**
-.007
-.016
(.013)
(.017)
(.016)
(.019)
-.018*
.053**
.002
-.013
(.012)
(.016)
(.015)
(.019)
.006
-.131**
.000
.033*
(.019)
(.025)
(.024)
(.029)
-.012
-.129**
-.015
.028*
(.018)
(.024)
(.023)
(.028)
-.017*
-.039**
.007
.015
(.014)
(.019)
(.018)
(.022)
-.022*
-.043**
.001
.013
(.014)
(.019)
(.018)
(.021)
.021*
-.007
.016*
.070**
(.013)
(.015)
(.014)
(.019)
.018*
-.005
.011
.047**
(.012)
(.014)
(.014)
(.017)
.022*
-.010
-.006
.006
(.011)
(.012)
(.012)
(.015)
.021*
-.009
-.008
-.002
(.011)
(.012)
(.011)
(.015)
.020*
-.014
-.034*
-.114**
(.016)
(.018)
(.018)
(.023)
.021*
-.013
-.027*
-.091**
(.015)
(.017)
(.017)
(.022)
.014*
-.006
-.012
-.030*
(.012)
(.013)
(.013)
(.017)
.014*
-.006
-.010
-.025*
(.012)
(.013)
(.013)
(.017)
30
Table 5: Share estimation results (cont)
Credit card
Specification
1
2
Debit card
Specification
3
4
1
2
3
4
Demographic
Education
High school
Graduate school
-.034**
.081**
(.013)
(.014)
-.025**
.057**
(.012)
(.013)
-.016*
.035**
(.011)
(.011)
-.013*
.026**
(.010)
(.011)
-.015
-.074**
(.016)
(.017)
-.021*
-.052**
(.015)
(.016)
-.011
-.018*
(.012)
(.013)
-.012*
-.012
(.012)
(.013)
.000
.032**
.104**
(.012)
(.012)
(.016)
-.010
.011
.062**
(.012)
(.011)
(.015)
-.005
.010*
.052**
(.010)
(.010)
(.013)
-.010*
.002
.034**
(.010)
(.010)
(.013)
.010
-.015
-.066**
(.015)
(.015)
(.020)
.017*
.001
-.033*
(.015)
(.014)
(.019)
.001
-.002
-.027*
(.011)
(.011)
(.015)
.003
.003
-.018*
(.011)
(.011)
(.015)
.061**
-.013
.010
-.026*
-.016
.005
.005
-.029*
(.024)
(.018)
(.015)
(.024)
(.018)
(.021)
(.021)
(.016)
.047**
-.000
.027*
-.021
.001
.011
.006
-.020*
(.022)
(.017)
(.014)
(.022)
(.017)
(.019)
(.020)
(.016)
.056**
.015*
.020*
.001
.004
-.004
.017
.004
(.019)
(.014)
(.012)
(.019)
(.015)
(.017)
(.017)
(.014)
.049**
.020*
.028**
.001
.010
-.001
.015
.005
(.019)
(.014)
(.012)
(.019)
(.015)
(.016)
(.017)
(.013)
-.009
-.005
.048**
.049*
.071**
.004
.050*
.084**
(.029)
(.022)
(.019)
(.029)
(.022)
(.025)
(.026)
(.020)
.005
-.024*
.031*
.047*
.057**
-.002
.051**
.077**
(.028)
(.021)
(.018)
(.028)
(.022)
(.024)
(.025)
(.019)
-.011
-.023*
.028**
.033*
.045**
.022*
.029*
.030**
(.022)
(.016)
(.014)
(.022)
(.017)
(.019)
(.019)
(.015)
-.006
-.030*
.022*
.035*
.040**
.019*
.030*
.030*
(.022)
(.016)
(.014)
(.022)
(.017)
(.019)
(.019)
(.015)
.021*
.022*
-.010*
(.011)
(.015)
(.010)
.017*
.011
-.013*
(.011)
(.014)
(.010)
.030**
.024*
.002
(.009)
(.012)
(.009)
.027**
.020*
.000
(.009)
(.012)
(.008)
.049**
.008
.067**
(.014)
(.018)
(.013)
.051**
.017
.066**
(.013)
(.018)
(.012)
.023**
.006
.015*
(.010)
(.014)
(.009)
.025**
.010
.016*
(.010)
(.014)
(.010)
Intercept
.139**
(.023)
.105**
(.023)
.286**
(.037)
.255**
(.037)
.259**
(.028)
.290**
(.028)
.201**
(.041)
.211**
(.042)
R-square
Adj. R-square
.1188
.1076
Financial (Income)
$40,000 - $59,999
$60,000 - $99,999
$100,000 and over
Census division
New England
Mid Atlantic
South Atlantic
ES Central
EN Central
WN Central
Mountain
Pacific
Technology
adoption
Direct deposit
Internet at home
Online banking
.2207
.2095
.4209
.4090
.4494
.4372
.1008
.0893
Notes: N=2000. Standard errors are in parentheses. ** t-value is greater than 2. * t-value is greater than 1.
31
.1692
.1573
.5050
.4948
.5149
.5041
Table 5: Share estimation results (cont)
Paper-based
Specification
1
2
3
4
Card related
No balance on credit
Rewards on debit
Rewards on credit
.007
-.056**
-.032**
(.011)
(.016)
(.012)
-.016*
-.010
-.037**
(.009)
(.014)
(.011)
Perceptions
Comfortable
Credit card
Debit card
Cash or checks
Fast
Credit card
Debit card
Cash or checks
Convenient
Credit card
Debit card
Cash or checks
Help me budget/spend
Credit card
Debit card
Cash or checks
Control over money
Credit card
Debit card
Cash or checks
Demographic
Female
Race
African American
Asian
Hispanic
Other
Age
18-34
45-54
55-64
65 and over
-.015**
-.018**
.042**
(.004)
(.005)
(.007)
-.014**
-.019**
.041**
(.004)
(.005)
(.007)
-.014**
-.012**
.015**
(.004)
(.005)
(.006)
-.014**
-.011**
.016**
(.004)
(.005)
(.006)
-.007
-.041**
.072**
(.011)
(.013)
(.014)
-.005
-.042**
.070**
(.011)
(.013)
(.014)
-.012
-.042**
.030**
(.014)
(.012)
(.012)
-.011
-.041**
.031**
(.014)
(.012)
(.012)
-.029**
-.032**
.037**
(.013)
(.012)
(.013)
-.022*
-.036**
.033**
(.013)
(.012)
(.013)
-.038**
(.011)
-.039**
(.011)
-.030**
(.009)
-.030**
(.009)
.056**
.006
.003
-.003
(.017)
(.022)
(.021)
(.026)
.057**
.008
.001
-.006
(.017)
(.022)
(.021)
(.026)
.056**
.000
.014
.001
(.015)
(.020)
(.019)
(.023)
.055**
-.002
.010
-.001
(.015)
(.020)
(.019)
(.023)
-.045**
.026*
.027*
.052**
(.015)
(.016)
(.016)
(.021)
-.042**
.023*
.024*
.051**
(.015)
(.016)
(.016)
(.021)
-.040**
.022*
.028**
.033*
(.013)
(.014)
(.014)
(.018)
-.037**
.020*
.028**
.036*
(.013)
(.014)
(.014)
(.018)
32
Table 5: Share estimation results (cont)
Paper-based
Specification
1
2
3
4
Demographic
Education
High school
Graduate school
.055**
-.015
(.014)
(.015)
.052**
-.012
(.014)
(.015)
.034**
-.023*
(.012)
(.013)
.031**
-.019*
(.012)
(.013)
-.007
-.020*
-.039**
(.014)
(.013)
(.018)
-.003
-.014*
-.028*
(.014)
(.014)
(.018)
.007
-.012*
-.024*
(.012)
(.012)
(.016)
.010
-.007
-.014
(.012)
(.012)
(.016)
-.046*
.027*
-.051**
-.003
-.059**
.000
-.054**
-.052**
(.026)
(.020)
(.017)
(.026)
(.020)
(.023)
(.023)
(.018)
-.044*
.034*
-.053**
-.007
-.062**
.000
-.057**
-.053**
(.026)
(.019)
(.017)
(.026)
(.020)
(.023)
(.023)
(.018)
-.038*
.017*
-.041**
-.015
-.052**
-.007
-.044**
-.031*
(.023)
(.017)
(.015)
(.023)
(.018)
(.020)
(.020)
(.016)
-.036*
.019*
-.045**
-.017
-.055**
-.008
-.044**
-.032**
(.023)
(.017)
(.015)
(.023)
(.018)
(.020)
(.020)
(.016)
-.073**
-.025*
-.060**
(.013)
(.017)
(.011)
-.070**
-.023*
-.055**
(.013)
(.017)
(.011)
-.056**
-.025*
-.020**
(.011)
(.015)
(.010)
-.054**
-.024*
-.019*
(.011)
(.014)
(.010)
Intercept
.559**
(.026)
.563**
(.027)
.444**
(.043)
.464**
(.044)
R-square
Adj. R-square
.1040
.0925
Financial (Income)
$40,000 - $59,999
$60,000 - $99,999
$100,000 and over
Census division
New England
Mid Atlantic
South Atlantic
ES Central
EN Central
WN Central
Mountain
Pacific
Technology
adoption
Direct deposit
Internet at home
Online banking
.1168
.1041
.3431
.3296
.3493
.3348
Notes: N=2000. Standard errors are in parentheses. ** t-value is greater than 2. * t-value is greater than 1.
33
Appendix: Multinomial logit model for payment choice by retail type
Retail type
Department stores
PIN debit
Signature debit
Credit
Check
Credit
Fast food restaurant
PIN debit
Signature debit
Check
Card related
No balance on credit
Rewards on credit
Reward on PIN
Reward on Sig
Demographic
Female
Race
Asian
Other
Age
Age^2
Education
Education^2
Income
Income^2
Technology
Direct deposit
Online banking
Intercept
-.049
.720**
.539**
.614**
.125
.145
.035
.245
.227
.287
.242
.821**
.204
.389
.100
.295
.317*
.204
.112
.165
.255*
.234
.277*
.257
-.571**
.253
-.034
.219
-.400*
.237
-.485
.617
.889*
-1.12**
-.022
.000
1.57**
-.168*
.056
.001
.582
.225
.044
.001
.679
.104
.094
.007
.118
-.714**
.042
-.001
1.36*
-.161*
.027
.000
.669
.237
.049
.001
.733
.112
.097
.006
.326
-.980**
.071*
-.001*
.887*
-.063
-.004
.001
.660
.253
.056
.001
.807
.123
.110
.007
.536
-.939**
.135**
-.001
.690
-.065
-.007
.002
.710
.316
.060
.001
.792
.121
.115
.008
1.11**
.469*
-.055*
.000
-.288
.069
-.114*
.005*
.343
.297
.046
.001
.770
.106
.078
.004
.156
.744**
-.071*
.001
1.33*
-.203*
.271*
-.015*
.537
.247
.053
.001
.703
.105
.136
.009
-1.30*
.481*
-.058*
.000
1.91*
-.217*
.068
-.004
.800
.254
.058
.001
1.02
.145
.094
.005
.914
-.841
.079
-.001
3.24
-.543
.216
-.104
2.28
1.05
.168
.002
4.28
.753
.978
.157
.445*
.334*
.229
.205
.378*
.495**
.246
.226
.748**
.487**
.270
.240
.384*
.200
.293
.267
.386*
.218
.302
.255
.173
.182
.288
.249
.591*
-.062
.347
.256
.509
-.575
1.02
.691
-1.10
1.43
-2.06*
1.60
-2.19*
1.76
-2.86*
1.78
.683
1.76
-3.35*
1.76
-4.02*
2.14
-7.54
8.24
Perceptions
Comfortable
Fast
Convenient
Easy to use
Preferred by stores
Safe
Taken right away
Help me budget/spend
For small amounts
Control over money
Easy to get refund
.585**
.281**
.511**
.429**
.318**
.175**
-.115*
.232**
.020
.565**
.385**
Log-likelihood
-1637.62
N
Standard errors are in parentheses.
.341**
.223**
.466**
.199*
.158*
.110*
-.134*
.241**
.706**
.197*
.207*
.055
.056
.129
.131
.095
.086
.100
.097
.102
.105
.087
-1010.97
1798
**
.063
.065
.168
.186
.120
.102
.122
.116
.108
.121
.114
1813
*
t-value is greater than 2. t-value is greater than 1
34
Retail type
Credit
Discount stores
PIN debit
Signature debit
Check
Credit
Drug stores
PIN debit
Signature debit
Check
Card related
No balance on credit
Rewards on credit
Reward on PIN
Reward on Sig
Demographic
Female
Race
Asian
Other
Age
Age^2
Education
Education^2
Income
Income^2
Technology
Direct deposit
Online banking
Intercept
Perceptions
Comfortable
Fast
Convenient
Easy to use
Preferred by stores
Safe
Taken right away
Help me budget/spend
For small amounts
Control over money
Easy to get refund
Log-likelihood
N
-.009
.516**
.593**
.656**
.134
.145
-.241
.156
.164
.246
.233
.656**
.207
.237
.808**
.207
-.226*
.168
-.061
.168
-.106
.212
.620**
.203
.139
.170
.390**
.151
.367*
.185
.583**
.204
1.03**
-.949**
-.014
.000
.534
-.053
-.047
.005*
.426
.209
.036
.000
.535
.078
.069
.004
.412
-.729**
.012
-.000
1.25**
-.140*
-.008
.003
.480
.191
.036
.000
.554
.082
.069
.004
-.015
-.695**
.066*
-.001*
1.11*
-.113*
-.136*
.008*
.637
.243
.054
.001
.739
.107
.083
.005
.646*
-1.20**
.120**
-.001**
1.02*
-.131
-.004
.005
.554
.287
.048
.001
.663
.097
.081
.005
.546*
-.189
-.110**
.001**
.610*
-.040
.101*
-.004*
.335
.220
.037
.000
.592
.083
.062
.003
-.608*
-.270*
-.079**
.001*
.921*
-.103*
.139**
-.006*
.384
.178
.034
.000
.508
.075
.057
.003
-.890*
-.483**
-.020
.000
.957*
-.100*
.110*
-.006*
.532
.222
.045
.001
.636
.093
.072
.004
-.902*
-.886**
.049
-.001
.219
-.009
.136*
-.009*
.740
.294
.053
.001
.676
.100
.087
.006
.388*
.362**
.211
.172
.633**
.249*
.205
.172
.636**
.367*
.264
.219
.491*
.455**
.248
.210
.278*
.199*
.210
.174
.350*
.104
.183
.156
.681**
.223*
.235
.198
.103
-.190
.245
.214
.155
1.18
-1.77*
1.23
-2.89*
1.71
-3.92**
1.55
-.055
1.35
.204
1.17
-1.88*
1.42
-1.26
1.72
.548**
.161**
.541**
.288**
.319**
.158*
.020
.301**
.143*
.407**
.169**
.650**
.203**
.667**
.395**
.309**
.243*
-.031
.158*
.316**
.398**
.274*
.053
.050
.120
.120
.088
.075
.090
.085
.095
.090
.080
.060
.049
.120
.125
.082
.070
.086
.084
.089
.086
.076
-1855.18
-1856.88
1761
1846
Standard errors are in parentheses. ** t-value is greater than 2. * t-value is greater than 1
35